Receptor residence time trumps drug-likeness and oral bioavailability in determining efficacy of complement C5a antagonists

Abstract

Drug discovery and translation are normally based on optimizing efficacy by increasing receptor affinity, functional potency, drug-likeness (rule-of-five compliance) and oral bioavailability. Here we demonstrate that residence time of a compound on its receptor has an overriding influence on efficacy, exemplified for antagonists of inflammatory protein complement C5a that activates immune cells and promotes disease. Three equipotent antagonists (3D53, W54011, JJ47) of inflammatory responses to C5a (3 nM) were compared for drug-likeness, receptor affinity and antagonist potency in human macrophages, and anti-inflammatory efficacy in rats. Only the least drug-like antagonist (3D53) maintained potency in cells against higher C5a concentrations and had a much longer duration of action (t1/2 ~ 20 h) than W54011 or JJ47 (t1/2 ~ 1 -3 h) in inhibiting macrophage responses. The unusually long residence time of 3D53 on its receptor was mechanistically probed by molecular dynamics simulations, which revealed long-lasting interactions that trap the antagonist within the receptor. Despite negligible oral bioavailability, 3D53 was much more orally efficacious than W54011 or JJ47 in preventing repeated agonist insults to induce rat paw oedema over 24 h. Thus, residence time on a receptor can trump drug-likeness in determining efficacy, even oral efficacy, of pharmacological agents.

Figure 1. Comparative properties and in vitro potencies of C5aR antagonists.

Top: Chemical structures for antagonists 3D53, W54011 and JJ47. Bottom: Properties and antagonist potencies of the three compounds. ^aMW = molecular weight, HBD = hydrogen bond donors, HBA = hydrogen bond acceptors, ClogP = calculated octanol-water partition coefficient, CLogS = calculated aqueous solubility, PSA = Polar surface area. ^bInhibition of Ca²⁺ release in different cells, under different conditions and against different concentrations of C5a. ^cVersus 100 nM rhC5a on neutrophils. ^dVersus 0.1nM rhC5a on neutrophils. ^eVersus 1.5 nM rhC5a on U937 cells.

Figure 2. Different mechanisms of C5aR antagonism by 3D53, W54011 and JJ47 against human C5a on human monocyte-derived macrophages.

Top row: Binding affinities of (A) 3D53, (B) W54011 and (C) JJ47 measured by displacement of [¹²⁵I]-C5a (25 pM) from HMDM. Middle row: Concentration dependent responses to C5a following treatment with antagonist (D) 3D53, (E) W54011 and (F) JJ47 at various concentrations (0 nM, ●; 3 nM, ■; 10 nM, ▲; 30 nM, ◆; 100 nM, 300 nM, ▼; 1000 nM, ★) with C5a (300 nM) as 100% response on human macrophages in a calcium release assay. Bottom row: Schild plots for antagonists (G) 3D53, (H) W54011 and (I) JJ47 against rhC5a. Calculated pA₂ values are 8.3 (3D53), 8.6 (W54011) and 8.3 (JJ47). Error bars are means ± SEM of three independent experiments (n = 3).

Figure 3. Dependence of antagonist potencies (IC₅₀) on C5a concentration in human monocyte-derived macrophages.

Inhibitory responses of antagonist (A) 3D53, (B) W54011 and (C) JJ47 against various concentrations of C5a (300 nM, ●; 100 nM, ■; 30 nM, ▲; 10 nM, ◆; 3 nM, ▼; 1 nM, ★) relative to C5a (300 nM) inducing calcium release in HMDM (100%). Calculated pIC₅₀ ± SEM and IC₅₀ values for the three C5aR antagonists are summarized in the table versus concentration of C5a. Error bars are means ± SEM of three independent experiments (n = 3).

Figure 4. 3D53 is more effective than W54011 and JJ47 in inhibiting C5a-induced migration or calcium release for human HMDM due to longer residence time on C5aR.

(A) Migration of HMDM through 5 μm Transwell inserts with or without 3 nM C5a for 16 h. For antagonism of C5aR, cells were treated with 3D53 (0.1 or 1 μM), W54011 (0.1 or 1 μM) or JJ47 (0.1 or 1 μM) for 1 h. Unbound antagonists were washed away prior to stimulation with 3 nM C5a. Chemotactic index is expressed as fold change against control. (B) C5a induced Ca²⁺ release from HMDM, pre-treated with 1 μM C5aR antagonist (●) 3D53, (▲) W54011 and (○) JJ47 for 1 h. Excess unbound antagonists were removed by washing. Antagonist residence time on C5aR was determined by subjecting treated cells to 3 nM C5a during the stated duration post antagonist-treatment up to 4 h and (C) up to 60 h for 3D53 in a calcium release assay. Calculated half-life for 3D53, W54011 and JJ47 are 18.2 h, 1.2 h and 0.6 h respectively. Error bars are means ± SEM of three independent experiments (n = 3). ***p < 0.001 by student t-test.

Figure 5. Antagonism of C5a-induced gene expression in human and rat macrophages by 3D53, W54011 and JJ47.

(A,B) HMDM were stimulated with 10 nM C5a and lysed after 30 min. For antagonism of C5aR, HMDM were pretreated with 1 μM 3D53, W54011 or JJ47 for 1 h. Excess unbound antagonists were removed by washing and subjected to 10 nM C5a stimulation after (A) 1 h or (B) 16 h post-treatment with antagonist. (C) Rat macrophages were stimulated with 0.1 μM C5a and cells are lysed after 30 min. For antagonism of C5aR, cells were pretreated with 1 μM 3D53, W54011 or JJ47 for 1 h. Excess unbound antagonists were removed by washing and subjected to C5a treatment after 1 h post-treatment with antagonist. Gene expression was measured by real-time PCR, normalized against 18S and converted to fold change relative to control. Error bars are means ± SEM of three independent experiments (n = 3). *p < 0.05, **p < 0.01 and ***p < 0.001 by student t-test.

Figure 6. Representation of C5aR-3D53 interactions at multiple time points in MD simulations.

C5aR-3D53 complex at (A) 10 ns: Arg6 of 3D53 faced towards ECL2 and made a short-lived contact with the Cys188 backbone carbonyl oxygen. (B) 50 ns: Arg6 underwent a large shift involving re-location and formation of an electrostatic interaction with Asp282, accompanied by movement of the top region of TM2 to shorten the distance to TM7. (C) 80 ns: similar protein-ligand interactions as at 50 ns, reflecting stable and long-lasting interactions between 3D53 and C5aR. Hydrogen bonds are depicted as dotted lines. (D) Distance measurements between stated residues and TM helices throughout MD simulations. (E) Root mean square fluctuation (RMSF) of C5aR residue sidechains throughout MD simulations.

Figure 7. 3D53, W54011 and JJ47 are orally active anti-inflammatory compounds that inhibit C5a agonist-induced paw oedema in male Wistar rats.

Paw oedema was induced in male Wistar rats by intraplantar administration of C5aR-PA (350 μg per paw in 100 μL of saline control). Paw swelling (% area change from baseline) was recorded at indicated timepoints. To determine comparative in vivo residence times, (A) 3D53 (5 mg/kg, p.o., n = 3 per group), (B) W54011 (30 mg/kg, p.o., n = 2 per group) and (C) JJ47 (30 mg/kg, p.o., n = 2 per group) were given orally 1 h prior to injection of C5aR-PA at specific time-points post-antagonism. Oral bioavailabilities of (D) 3D53 (F ~ 2%), (E) W54011 (F ~ 74%) and (F) JJ47 (F ~ 12%) were measured using LCMS to quantify plasma concentrations. C5aR antagonist was administering as an intravenous dose (1 mg/kg via an implanted jugular vein catheter in 50:50 DMSO/saline) versus an oral dose (10 mg/kg in olive oil, via gavage). Plasma was collected at various time points and analyzed. Error bars are means ± SEM normalized to maximal swelling of saline control. *p < 0.05, **p < 0.01 and ***p < 0.001 by student t-test.